[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: [PATCH] optionally use $SVN_USERNAME as login

From: Branko ─îibej <brane_at_xbc.nu>
Date: 2005-07-25 02:36:16 CEST

Mark Benedetto King wrote:

>On Sat, Jul 23, 2005 at 05:38:08PM +0200, Branko ??ibej wrote:
>
>
>>kfogel@collab.net wrote:
>>
>>
>>
>>>kfogel@collab.net writes:
>>>
>>>
>>>
>>>
>>>>I'm not sure I agree. People know about setting environment
>>>>variables, plus CVS has $CVS_USER. Aliasing 'svn' to a complex
>>>>command involving backticks is much less intuitive. And, as Ben
>>>>pointed out, it won't work for commands that don't accept
>>>>'--username'.
>>>>
>>>>
>>>>
>>>>
>>>(Excuse me, I just realized I was wrong about CVS_USER -- that's
>>>something totally different.)
>>>
>>>I still think SVN_USERNAME or SVN_USER is a good idea, for the reasons
>>>already given, though.
>>>
>>>
>>>
>>>
>>I explained in a previous post that Subversion should be caching the
>>username associated with a repository. If that doesn't work, we should
>>fix the credentials caching, not introduce dependencies on environment vars.
>>
>>Consider this a -1 until someone shows that either a) credentials
>>caching or b) teaching svn to always accept --username, aren't
>>sufficient for solving this particular contrived problem.
>>
>>
>>
>
>Before you sent this message, I committed a semantically similar change
>to the one being discussed:
>
>r15393 | mbk | 2005-07-22 17:09:50 -0400 (Fri, 22 Jul 2005) | 7 lines
>
>I had percieved consensus where (it is now apparent) there was none.
>
>I don't want to bypass your veto by fiat; I will happily revert the
>change if the following argument does not satisfy you:
>
>1.) Credentials caching is insufficient.
>
>If we ignore the "first-use" case (where caching obviously loses) on
>the grounds that it happens infrequently in practice, we are still
>left with the fact that (except on Windows) many people quite reasonably
>consider our plain-text credentials caching unacceptable.
>
>
Note that I said "credentials", not "password". Credentials include
usernames. If we're currently unable to tell the client to cache the
username but not the password, then by all means let's teach it to do so.

(Oh, and people don't consider plain-text password caching acceptable on
Windows, either -- that's why Subversion encrypts passwords on Windows
as of 1.2.0.)

>2.) Always accepting --username is insufficient.
>
>Okay, I can't argue with this because it /would/ be sufficient, but what
>I *can* say is that always accepting --username is not necessarily the
>right thing to do. Should we also always accept --editor-cmd? No,
>because we have $SVN_EDITOR...
>
>
And I do consider $SVN_EDITOR to be a mistake. Note that we also
understand $EDITOR and $VISUAL, and that's fine, because they are
considered standard ways of defining the default editor on Unix.

I'd be much obliged if you'd revert r15393 until we setlle this issue.

-- Brane

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Mon Jul 25 02:35:54 2005

This is an archived mail posted to the Subversion Dev mailing list.